This is an unblinded, voluntary, open-label toxicity trial assessing zidovudine as a candidate agent for post-exposure chemoprophylaxis for healthcare workers (HCW) sustaining occupational exposures to HIV. Because of increasing concerns about zidovudine-resistant strains of HIV, HCW whose source patients have taken equal to or over 6 months of zidovudine therapy may elect to add didanosine as a second agent. The two secondary purposes of this study are: l) to assess the effect of zidovudine or zidovudine plus didanosine chemoprophylaxis on the temporal sequence of the appearance of markers of HIV infection should it occur in a HCW(s) taking chemoprophylaxis; and 2) to describe the epidemiology of exposures to HIV for which chemoprophylaxis is elected in institutions participating in this study. Zidovudine is initiated as soon as possible for exposed HCW, but no longer than 72 hours post-exposure, at a dose of 200 mg 6x/day for 3 days, followed by 200 mg 3x/day for 25 days; total treatment course is 28 days. Didanosine can be added as a second agent at any time up to 72 hours after the first dose of zidovudine. The didanosine dosage is 400 mg/day of chewable tablets, and the total duration of prophylaxis remains at 28 days. Exposed HCW are followed for a minimum of 12 months. After the baseline enrollment visit, follow-up visits occur at 2, 4, and 6 weeks, and 3, 6, and 12 months. Individuals are assessed for drug toxicity as well as for early signs of HIV infection. Individual HCW who have elected to take zidovudine have, in general, tolerated the drug well. Subjective toxicities occur in the majority of HCW; nonetheless, symptomatology has not correlated at all with objective hematologic toxicities. No investigator has discontinued any course of zidovudine because of laboratory toxicity in a HCW. One HCW has developed evidence of HIV infection following exposure despite administration of zidovudine. Three HCW have taken zidovudine and didanosine together; all 3 discontinued prophylaxis early because of subjective symptomatology.